RE: [PATCH v4 2/6] KVM: nVMX: Enable nested virtualize x2apic mode

From: "Zhang, Yang Z" <yang.z.zhang@intel.com>
To: Wincy Van <fanwenyi0529@gmail.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>,
	"gleb@kernel.org" <gleb@kernel.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"kvm@vger.kernel.org" <kvm@vger.kernel.org>,
	Wanpeng Li <wanpeng.li@linux.intel.com>,
	Jan Kiszka <jan.kiszka@web.de>
Subject: RE: [PATCH v4 2/6] KVM: nVMX: Enable nested virtualize x2apic mode
Date: Thu, 29 Jan 2015 04:05:29 +0000	[thread overview]
Message-ID: <A9667DDFB95DB7438FA9D7D576C3D87E0AC2E3F5@SHSMSX104.ccr.corp.intel.com> (raw)
In-Reply-To: <CACzj_yXbEirXExiC3p2m_zf9AGLYGKU5fLGHdizPN10eL4Q5Ug@mail.gmail.com>

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain; charset="utf-8", Size: 2916 bytes --]

Wincy Van wrote on 2015-01-29:
> On Thu, Jan 29, 2015 at 10:54 AM, Zhang, Yang Z <yang.z.zhang@intel.com>
> wrote:
>>> -8646,7 +8750,8 @@ static void prepare_vmcs02(struct kvm_vcpu
>>> *vcpu, struct vmcs12 *vmcs12)
>>>                         else
>>> vmcs_write64(APIC_ACCESS_ADDR,
>>> 
>>> page_to_phys(vmx->nested.apic_access_page));
>>> -               } else if
>>> (vm_need_virtualize_apic_accesses(vmx->vcpu.kvm)) {
>>> +               } else if
>>> + (!(nested_cpu_has_virt_x2apic_mode(vmcs12))
>>> + &&
>>> +
>>> + (vm_need_virtualize_apic_accesses(vmx->vcpu.kvm))) {
>>>                         exec_control |=
>> 
>> You don't load L2's apic_page in your patch correctly when x2apic
>> mode is
> used. Here is the right change for prepare_vmcs02()(maybe other place
> also need change too):
>> 
>> @@ -8585,7 +8585,8 @@ static void prepare_vmcs02(struct kvm_vcpu *vcpu,
>> struct vmcs12 *vmcs12)
>> 
> CPU_BASED_ACTIVATE_SECONDARY_CONTROLS))
>>                         exec_control |=
>> vmcs12->secondary_vm_exec_control;
>> 
>> -               if (exec_control &
>> SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES) { +               if
>> (exec_control & (SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES | + +
>> SECONDARY_EXEC_VIRTUALIZE_X2APIC_MODE)) {
>>                         /*
>>                          * If translation failed, no matter: This
>> feature
> asks
>>                          * to exit when accessing the given address,
>> and if it @@ -8594,7 +8595,8 @@ static void prepare_vmcs02(struct
> kvm_vcpu *vcpu, struct vmcs12 *vmcs12)
>>                          */
>>                         if (!vmx->nested.apic_access_page)
>>                                 exec_control &=
>> - ~SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES; +                          
>>       ~ (SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES | + +
>> SECONDARY_EXEC_VIRTUALIZE_X2APIC_MODE);
>>                         else
>>                                 vmcs_write64(APIC_ACCESS_ADDR,
>> page_to_phys(vmx->nested.apic_access_page));
>> 
> 
> I think we don't need to do this, if L1 enables x2apic mode, we have
> already checked that the vmcs12->SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES
> is 0. "exec_control |= vmcs12->secondary_vm_exec_control;" merged L1's
> settings, including x2apic mode. the special case is
> vm_need_virtualize_apic_accesses, if vm_need_virtualize_apic_accesses
> returns true, the nested_cpu_has_virt_x2apic_mode will prevent us to set
> the apic access bit.

I just realized that we are talking different thing. I am thinking about VIRTUAL_APIC_PAGE_ADDR(my mistake :)). And it has been handled correctly already. For APIC_ACCESS_ADDR, you are right. The current implementation can handle it well.

> 
> 
> Thanks,
> Wincy

Best regards,
Yang

ÿôèº{.nÇ+‰·Ÿ®‰†+%ŠËÿ±éÝ¶\x17¥Šwÿº{.nÇ+‰·¥Š{±þG«éÿŠ{ayº\x1dÊ‡Ú™ë,j\a¢f£¢·hšïêÿ‘êçz_è®\x03(éšŽŠÝ¢j"ú\x1a¶^[m§ÿÿ¾\a«þG«éÿ¢¸?™¨èÚ&£ø§~á¶iO•æ¬z·švØ^\x14\x04\x1a¶^[m§ÿÿÃ\fÿ¶ìÿ¢¸?–I¥